Morphological characterization of vanadium oxide supported on carbon-coated monoliths using AFM

Morphological characterization of vanadium oxide supported on carbon-coated monoliths using AFM

Applied Surface Science 228 (2004) 135–142 Morphological characterization of vanadium oxide supported on carbon-coated monoliths using AFM E. Garcı´a...

320KB Sizes 0 Downloads 24 Views

Applied Surface Science 228 (2004) 135–142

Morphological characterization of vanadium oxide supported on carbon-coated monoliths using AFM E. Garcı´a-Bordeje´a,*, M.J. La´zaroa, R. Molinera, J.F. Galindob,1, J. Sotresc,1, A.M. Baro´b,1 a

Instituto de Carboquı´mica (CSIC), Energia y medioambiente, Miguel Luesma Casta´n 4, 50015-Zaragoza, Spain b Dep. Fı´sica de la Materia Condensada C-III, Universidad Auto´noma de Madrid, 28049-Madrid, Spain c Nanotec Electro´nica S.L., Pabello´n C, Campus Cantoblanco, 28049-Madrid, Spain Received 2 December 2003; received in revised form 2 January 2004; accepted 2 January 2004

Abstract Vanadium has been supported on carbon-coated monoliths with three different loadings, viz. 4, 6 and 8 wt.%. The structure of vanadium catalyst has been characterized by atomic force microscopy, N2 adsorption and XRD. All these techniques agree to indicate that vanadium is dispersed as overlayers up to 6 wt.% loadings. For higher loadings crystallites appear. The size and shape of these crystallites (12 nm average size) have been characterized by AFM. The amount of vanadium per gram of support deposited as monolayer on our carbon support can be higher than in a traditional oxide supports as, e.g. TiO2. This is a consequence of the higher surface area of carbon. The monolayer dispersion of vanadium on the carbon-coated monoliths endows this catalyst with a great potentiality to be used in reactions as, e.g. the SCR of NO at low temperature. # 2004 Elsevier B.V. All rights reserved. Keywords: Atomic force microscopy; Vanadium oxide; Carbon support; Catalysis

1. Introduction Due to the unique redox and acid base properties of vanadium oxide, V-containing catalysts have been extensively used in a large number of catalytic processes involving selective oxidation reactions (such as conversion of butane to maleic anhydride, oxidation of o-xylene, 1,3-butadiene, methanol, CO, ammoxidation of hydrocarbons, selective catalytic reduction of NO and partial oxidation of methane) [1–4]. * Corresponding author. Tel.: þ34-976-733977; fax: þ34-976-733318. E-mail address: [email protected] (E. Garcı´a-Bordeje´). 1 On leave at the Instituto de Carboquı´mica (CSIC), Zaragoza.

The catalyst used in the industrial practice is normally supported on metal oxides, e.g. TiO2, Al2O3, ZrO2, SiO2. Supporting a metal oxide on the surface of another oxide was initially proposed to improve the catalytic activity of the active metal oxide phase due to a gain in surface area and mechanical strength. The molecular structures and morphologies of vanadium supported on these oxides have been intensively studied [5–10]. These studies have revealed that the geometric and electronic structure of surface vanadium oxides is the fundamental basis for the catalytic performance. There is the general consensus that structural and reactivity characteristics of vanadium oxide are modified with respect to bulk vanadium oxide by interaction with the oxide support surface

0169-4332/$ – see front matter # 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2004.01.001

136

E. Garcı´a-Bordeje´ et al. / Applied Surface Science 228 (2004) 135–142

and that this specific oxide–oxide interaction leads to a spreading of vanadium oxide over the support surface to form a thin monolayer film. The dispersion depends on the available surface area as well as the availability of surface hydroxyl groups. Hence, the surface population of hydroxyl sites determines the monolayer coverage of the supported oxide. In the preparation of these catalysts upon calcinations, the formed dehydrated vanadium oxides directly anchor to the surface via an esterification reaction with the hydroxyl groups of the support, resulting in the formation of surface vanadium oxides. Spectroscopic studies of VOx supported on different metal oxides showed the coexistence of V2O5 crystallites and species which have V–O–V bonds [8–10]. Four kinds of VOx species are proposed for supported VOx (Fig. 1): (a) isolated species (orthovanadate); (b) dimeric (pyrovanadate) or polymeric species, these polymeric species can also form chains of vanadium ions building up two-dimensional overlayers; (c) VOx threedimensional thin overlayers; (d) V2O5 crystallites. Some authors [9], assuming the presence of monovanadates (a) or polyvanadates (b), have estimated the theoretical monolayer coverages in TiO2 support. These theoretical monolayer coverages correspond to 0.8 and 2.4 vanadium atoms/nm2 for mono- and polyvanadates, respectively. For instance, Wachs and coworkers [2] prepared V supported on TiO2 with

different loadings, finding that monolayer coverage is achieved at 2.6 vanadium/nm2. This value agrees with the theoretical monolayer coverage for polyvanadates. But the term ‘‘monolayer’’ is not restricted to only the formation of single bi-dimensional-like sheet. ‘‘Monolayer’’ catalyst is also used to indicate samples in which the vanadium oxide characteristics are analogous to those of a single overlayer catalyst. This can take place up to three to five theoretical monolayers. In this sense, ‘‘monolayer’’ surface coverage of vanadia overlayer on oxide support was determined to be approximately 7–8 VOx per nm2 [1]. Hereinafter, this second concept of monolayer will be used, i.e. monolayer coverage as the maximum loading at which the supported oxide is deposited as surface species, free of three-dimensional aggregates. Much more scarce is the literature dealing with the characterization of the structures of vanadium supported on carbon support, in despite that this catalyst has shown recently very good performance for the selective catalytic reduction of NOx at low temperature [11–16]. To the best of our knowledge none of these authors has characterized the structure of vanadium on carbon. The limited use of carbon can be based on the general belief that the interaction of the active phase with carbon is weaker than with an oxide support, thus leading to lower dispersions. However, this is not completely true because the surface chem-

Fig. 1. Possible structures of vanadium described on oxide supports: (a) isolated species (orthovanadate); (b) polymeric species; (c) VOx overlayers; (d) V2O5 crystallites.

E. Garcı´a-Bordeje´ et al. / Applied Surface Science 228 (2004) 135–142

istry of activated carbon can be tuned to create hydroxyl groups and, moreover, the surface area is much higher than in the metal oxide supports. In this work, a novel catalytic system based on vanadium supported on carbon has been prepared. Vanadium has been supported on the carbon-coated monoliths with three different loadings. The structure of vanadium has been characterized by several techniques, viz. atomic force microscopy, N2 adsorption and XRD. The objective of this work is determining the loading at which vanadium is dispersed on our carbon as a monolayer and comparing with the monolayer coverage in other oxide supports.

2. Experimental 2.1. Preparation of carbon-coated monoliths Cordierite monoliths (62 cm2, 1 cm diameter, 5 cm length) were coated with a polymer blend by the method described elsewhere [16,17]. The polymers used are Furan resin (Huttenes-Albertus) and polyethylene glycol 6000 MW (Sigma–Aldrich). After curing in air at 313 K during 24 h followed by 24 h at 373 K, the monoliths were carbonized at 973 K. Subsequently, they were activated with CO2 at 1173 K and 1 bar, during 8 h (30% carbon burn-off) and treated 24 h with 2 N HNO3 at room temperature to create oxygenated surface groups. 2.2. Vanadium catalyst preparation The carbons prepared were loaded with different amounts of vanadium, i.e. 4, 6 and 8 wt.%. The impregnation was carried out by equilibrium adsorption of Ammonium metavanadate in the stoichiometric amount to get the desired Vanadium loading on carbon. To facilitate the dilution of ammonium metavanadate ca. 10 mg of oxalic acid was added. Under these conditions the pH of solution remains neutral and the solution holds a yellow color indicative of the presence of VO2þ. The monoliths placed in a holder were introduced in a vessel with 100 ml of the impregnating solution. The vessel has a stirrer at the bottom that creates a continuous flux of the vanadium solution through the monoliths. This guarantees that vanadium is deposited

137

inside the channels. The V content of the solutions before and after impregnation was analyzed by ICP. From the difference in V content of these solutions the V loading was calculated. Finally, the catalyst was calcined in Ar at 673 K. 2.3. Characterization of vanadium catalyst In order to characterize the topography of the samples we use atomic force microscopy. The images were recorded with a commercial system from Nanotec Electro´ nica S.L. The AFM was operated in the dynamic mode, where the cantilever is oscillated at its eigenfrequency (72 kHz) and feedback is adjusted in order that there is no contact between the sample and the tip (tapping mode). In addition to topography image, we also collected the phase shift image. Phase shift data can be collected simultaneously with height data. Phase shift detection measures the phase lag between the drive signal and the actual tip response signal. In phase shift detection the contrast mechanism is a convolution of topography and material/chemical properties of the surface. This allows distinguishing between the vanadium and carbon substrate due to their different chemical composition. The vanadium particle dimensions were determined by analysis of the height AFM images collected using the software supplied by WSxM. Section analysis provides a cross-sectional height profile of the surface along a reference line, which is drawn across the region of interest. This permits accurate estimation of the vertical dimensions of the surface features. Lateral dimensions of particles in this size range can be exaggerated by finite tip radius effects. More than 30 images of every sample were recorded some of them being featureless. The average number of vanadium particles used to estimate the average V particle diameter was approximately 25. Nitrogen adsorption was performed on a Quantachrome autosorb-1 at 77 K. The pore volume and pore size distribution were calculated using the BET method. From the physisorption measurements with nitrogen the specific surface area has been calculated by the BET (Brunauer, Emmet and Teller) theory in the relative pressure range 0.01–0.10 following standard ASTM-4365, which is applicable to microporous materials. The t-plot method was applied to calculate the micropore volume and micropore surface area.

138

E. Garcı´a-Bordeje´ et al. / Applied Surface Science 228 (2004) 135–142

Pore size distribution was obtained from the desorption branch of the N2 isotherm according to the BJH method. X-ray diffraction (XRD) patterns were recorded using a Philips PW 1840 diffractometer using nickel-filtered Cu Ka radiation.

3. Results In the preparation of supported catalysts by impregnation, the sorption of metal ions in solution is influenced the physico-chemical properties of the support, pH of impregnation and nature of the solvent. When the pH of impregnating solution is higher than the point of zero charge (PZC), the surface of the support will become negatively charged, and the adsorption of cations is favored [18]. The main goal of oxidation with HNO3 is to obtain a more hydrophilic surface with a relatively large number of oxygen containing surface groups, hence decreasing the PZC. In our case the PZC of the carbon treated with HNO3 has a value around 3 and the pH of the solution is higher (pH 7) that favors the adsorption of the cation (VO2þ) on the negatively charged oxygenated surface groups. This is confirmed by the rapid decoloration of the yellow vanadium solution. These groups are located in the edges and defects of the basal planes and in the micropores. Accordingly, the vanadium ions will be anchored on these sites. Furthermore, the groups created after the oxidation with HNO3 contribute to evenly disperse the vanadium along the length of the

monolith. The even distribution of vanadium along the monolith was verified by cutting the monolith in three parts and analyzing the vanadium content by ICP. The vanadium content in each part was very similar and equivalent to the nominal vanadium content. Fig. 2 shows a peculiar AFM image with four very regular features. Fig. 2a is the topographical image and Fig. 2b is the phase shift image. The different contrast between the features and the support in the phase shift image enables us to state that the features in the pictures have different chemical composition than the carbon support, i.e. these crystallites can be unambiguously ascribed to vanadium. Figs. 3–5 show some AFM pictures with the features characteristics of every sample. At the left there is a top view image and at the right there is the profile in nm along the line drawn in the top view. It is important to note that AFM is a technique than only allows seeing features in the external surface thereby it is blind to catalyst particles in the pores. Also crystallites smaller than 4 nm can not be discerned from the roughness of the support. No apparent crystallites were observed in any image of sample with 4 wt.% V (Fig. 3b). The images of sample with 4 wt.% resemble those of the carboncoated support (Fig. 3a). In sample with 6 wt.% V most of the images explored were featureless and the features found were more or less isolated. Fig. 4a and b correspond to two pictures of sample 6 wt.% vanadium with some crystallites which have the appearance usually found in this sample. The crystallite average height is 7 nm and the diameter between 60 and 90 nm (Table 1).

Fig. 2. AFM image showing four neighboring vanadium crystallites in sample 8 wt.%: (a) topographic image; (b) phase shift image that allows to distinguish the different chemical nature of the crystallites and the support.

E. Garcı´a-Bordeje´ et al. / Applied Surface Science 228 (2004) 135–142

Fig. 3. Standard AFM image of: (a) support, (b) sample with 4% vanadium. Left, AFM top view; right, profile along the line in the top view.

Fig. 5 illustrates two images representative of the sample with 8 wt.% vanadium. Comparing with the crystallites present in sample with 6 wt.% V, there is a significant increase both in height and in the lateral

Fig. 4. Pictures of samples with 6% of vanadium with some isolated crystallites. Left, AFM top view; right, profile along the line in the top view.

139

Fig. 5. Pictures of samples with 8% of vanadium with the typical crystallites observed in this sample. Left, AFM top view; right, profile along the line in the top view.

dimensions (Table 1). Furthermore, the surface of the catalyst is more densely decorated with crystallites than sample with 6 wt.% V. This is a consequence of the higher loading and also of the fact that in sample with 6 wt.% V most of the vanadium is not as crystallites but disperse. Crystallites found in both samples (6 and 8 wt.%) have some common features, viz. they are flat at the top and they are located at the entrance of macropores or depressions of the support (darkest color in the pictures). The flatness of the top agrees with the laminar structure of V2O5 reported in the literature [19,20]. The borders of pores correspond to the edges of carbon basal planes where there is a high density of oxygenated surface groups that act as anchoring sites for vanadium precursor. Therefore it is not surprising that the crystallites are located near the entrance of pores. Fig. 6 depicts the particle size distribution of samples with 6 and 8 wt.% vanadium in which it is apparent that the height of crystallites in sample 6 wt.% V is smaller than in sample 8 wt.% V. It is important to note that the abundance in the y-axis is relative to the total number of crystallites found by AFM, not to the total vanadium content. It is obvious that most of vanadium in sample with 6 wt.% V is not

E. Garcı´a-Bordeje´ et al. / Applied Surface Science 228 (2004) 135–142

140

Table 1 Parameters obtained by the treatment of images by digital software V loading (%)

Nominal VOx surface density (V nm2)

Average rms roughness (nm)

0 4 6 8

0 0.75 1.16 1.78

2.5 1.4 2.0 3.5

   

Fig. 6. Particle size distribution of crystallites detected by AFM of carbon-coated monoliths with: (&) 6 wt.% and ( ) 8 wt.% vanadium.

detectable by AFM because either it is inside the pores or it has height smaller than 4 nm or both. Also some average values of the crystallites obtained by the digital treatment of images are displayed in Table 1. This table also shows the nominal VOx surface density calculated taking into account the BET surface area in Table 2. As observed in Table 1 the average rms roughness increases with the vanadium loading that is a consequence of the increase of number and size of crystallites. It is also observed that the average rms roughness of the sample with 4 wt.% vanadium decreases with

0.05 0.03 0.04 0.05

Average crystallite height (nm)

Lateral dimension (nm)

– – 7 12

– – 60–90 130–170

respect to that of the support. A possible interpretation is that the vanadium oxide is deposited as a layer in the mesopore walls decreasing their diameter. This would render an apparently smoother surface because the micropores are inaccessible to the AFM tip. The carbon coating support has micro- and mesopores as illustrated in the pore size distribution (Fig. 7). In this figure it is apparent that there is a reduction in the mesopore volume upon impregnation of vanadium on the carbon support. Table 2 shows the values of the pore volume and surface area both of micro- and mesopores. Comparing the textural features of the samples loaded with different percentages of vanadium with those of the support, the total pore volume and area remain practically unchanged up to 6 wt.% loading. Hence, there is no pore clogging. The exception is sample loaded with 8 wt.% where there is a significant decrease of the surface area and pore volume both in the micro and mesoporous range. The decrease in surface area is higher in the case of microporosity. This occurs because the big crystallites detected by AFM plug some micro- and mesopores. In the case of the samples loaded with 4 and 6% of vanadium the total surface area and pore volume apparently do not change, but intriguingly the micropore volume and surface area increases in detriment of the mesopore volume and area. This fact can be tentatively explained by the coverage of the narrow

Table 2 Textural parameters by N2 adsorption V loading (%)

BET area (m2 g1)

Total pore volume (cc g1)

Micropore area (m2 g1)

Mesopore area (m2 g1)

Micropore volume (cc g1)

Mesopore volume (cc g1)

0 4 6 8

617 629 607 529

0.96 0.86 0.90 0.77

326 402 354 259

292 227 253 270

0.18 0.23 0.20 0.18

0.78 0.63 0.70 0.60

E. Garcı´a-Bordeje´ et al. / Applied Surface Science 228 (2004) 135–142

Fig. 7. Pore size distribution of carbon coating by N2 adsorption: (—) carbon support; (^) 4 wt.% V; (&) 6 wt.% V; (~) 8 wt.% V.

mesopores walls with a layer of vanadium. Hence these mesopores decrease their diameter becoming micropores. The results described above were also confirmed by XRD. The diffraction peaks of V2O5 were only found in catalyst with vanadium loading of 8 wt.% (not shown) indicating that the dispersion on this latter catalyst is poorer than in the former ones. It was not possible to derive the average crystallite size from XRD since the peaks are very small and broad.

141

other oxide supports. In despite of this, the surface area of our carbon (617 m2/g) is much higher than the surface area of other oxide supports. Assuming an anatase support with very high surface area, e.g. 100 m2/g, the total amount of vanadium loaded at monolayer coverage will be 1.16 mmol/g of support while in our carbon it is 1.19 mmol/g. This indicates that the total amount of active sites, supposing the active sites are surface vanadium species, on our carbon support is slightly higher than that on a traditional oxide support as TiO2. The more vanadium can be deposited in a support as monolayer, the less amount of catalyst will be needed for a certain process that entails reduction in cost of catalyst and energy. Furthermore, the achievement of higher vanadium loadings at monolayer coverage is envisaged as feasible in our carbon support. To this end, the surface area and also the number of hydroxyl surface groups of our carbon can be increased. This latter would enhance the interaction between carbon and vanadium oxide.

5. Conclusions 4. Discussion According to our results, in sample with 8 wt.% V both the crystallites found in AFM images, the diffraction peaks of V2O5 and the decrease of porosity measured by N2 adsorption indicate that the predominant species is V2O5 crystallites. Conversely, in sample with 4, 6 wt.% V no crystallites and very few crystallites were found by AFM, respectively. This points out that most of vanadium is in a dispersed fashion either as isolated species, polymeric chains or as overlayers. This is also supported by the fact that there is not pore plugging and that the mesopores reduce their diameter because layers of vanadium are covering their walls. Therefore, the monolayer coverage in our vanadium/carbon catalyst, defined as maximum loading at which vanadium is dispersed before three-dimensional aggregates start to grow, is around 6 wt.% vanadium. Accordingly, the nominal VOx surface density at monolayer coverage of our carbon is 1.16 V nm2 which is lower than in other oxide supports, viz. 7 V nm2 [1], likely because the interaction of vanadium with our carbon is weaker than with

AFM has revealed as a very useful technique for the characterization of catalyst crystallites in a relatively flat surface as that of a monolith channel. Vanadium is supported in our carbon-coated monoliths as V2O5 overlayers up to approx. 6 wt.%. Above this loading V2O5 crystallites appear. These findings are also supported by N2 adsorption and XRD. The nominal VOx surface density at monolayer coverage of our carbon is lower than in other traditional oxide supports due to the weaker interaction of vanadium with this carbon. Nevertheless, the amount of active phase per gram of support that our carbon can accommodate as monolayer is higher than in TiO2, i.e. the standard support for vanadium in some reactions as, e.g. the selective catalytic reduction of NOx. This is a consequence of the much higher surface area of our carbon support.

Acknowledgements The authors wish to thank Hu¨ ttenes-Albertus (Hannover) and Corning (New York) for supplying

142

E. Garcı´a-Bordeje´ et al. / Applied Surface Science 228 (2004) 135–142

the Furan resin and cordierite monoliths, respectively. The authors acknowledge the Spanish Ministry of Science and Technology (PPQ-2002-02698 and BFM-2001-0209) and ‘‘Diputacion General de Arago´ n’’ for the financial support. E. Garcı´a-Bordeje´ is also indebted to the MCyT for a ‘‘Ramo´ n y Cajal’’ contract. References [1] I.E. Wachs, B.M. Weckhuysen, Appl. Catal. A: Gen. 157 (1997) 67. [2] G. Deo, I.E. Wachs, J. Haber, Crit. Rev. Surf. Chem. 4 (1994) 141. [3] G. Bond, S.F. Tahir, Appl. Catal. 71 (1991) 1. [4] M.A. Ban˜ ares, Catal. Today 51 (1999) 319. [5] I.E. Wachs, G. Deo, B.M. Weckhuysen, A. Andreini, M.A. Vuurman, M. de Boer, M.D. Amiridis, J. Catal. 161 (1996) 211. [6] I.E. Wachs, Catal. Today 27 (1996) 437. [7] H. Eckert, I.E. Wachs, J. Phys. Chem. 93 (1989) 6796.

[8] B.M. Weckhuysen, D.E. Keller, Catal. Today 78 (2003) 25. [9] G. Centi, Appl. Catal. A: Gen. 147 (1996) 267. [10] K. Inumaru, M. Misono, T. Okuhara, Appl. Catal. A: Gen. 149 (1997) 133. [11] S. Kasaoka, E. Sasaoka, I. Iwasaki, Bull. Chem. Soc. Jpn. 62 (1989) 1226. [12] Z. Zhu, Z. Liu, H. Niu, S. Liu, J. Catal. 187 (1999) 245. [13] Z. Zhu, Z. Liu, S. Liu, H. Niu, T. Hu, T. Liu, Y. Sie, Appl. Catal. B: Environ. 26 (2000) 25. [14] Z. Zhu, Z. Liu, S. Liu, S.H. Niu, Appl. Catal. B: Environ. 30 (2001) 267. [15] T. Valdes-Solis, G. Marba´ n, A.B. Fuertes, Appl. Catal. B: Environ. 46 (2003) 261–271. [16] E. Garcı´a-Bordeje´ , L. Calvillo, M.J. La´ zaro, R. Moliner, Appl. Catal. B: Environ., in press. [17] E. Garcı´a-Bordeje´ , F. Kapteijn, J.A. Moulijn, Carbon 40 (2002) 1079. [18] L.R. Radovic, F. Rodriguez-Reinoso, in: Carbon Materials in Catalysis, Chemistry and Physics of Carbon, vol. 25, Dekker New York, 1997, p. 243. [19] J. Haber, M. Witko, R. Tokarz, Appl. Catal. A: Gen. 157 (1997) 3. [20] A. Da Costa, C. Mathieu, Y. Barbaux, H. Poelman, G. Dalmai-Vennik, L. Fiermans, Surf. Sci. 370 (1997) 339.